Measuring Progress in Carrier SDN
Six years on, is carrier SDN a success or a failure? It's not a simple question to answer. Perceptions are defined by one's expectations, and with early expectations for both SDN and NFV sky-high, failure to meet expectations was inevitable. But if we remove perceptions and expectations from the analysis, we get a more objective read on SDN's progress in carrier networks.
In our latest global operator survey results, recently published in Charting the Path to Network Automation & Disaggregation: Carrier SDN Survey Analysis, 35% of respondents reported that they are at some level of commercial SDN deployment, with the remaining 65% at various stages of trials and market education. Yes, there is a long way to go in both depth and breadth of SDN adoption, but the percentage reporting commercialization is double the figure reported in our 2016 carrier SDN survey. It may not be as much progress as some initially hoped for, but the SDN commercialization needle is moving.
Progress can also be observed in the relationship between SDN and NFV. When defining NFV, ETSI was clear in stating that NFV and SDN were complementary but different technologies. Still, their early histories were closely intertwined and it was difficult to identify much carrier SDN activity outside the virtual world. But Heavy Reading data illustrates that SDN is, in fact, advancing commercially, and often separately from NFV initiatives. In other words, while SDN is still often coupled with NFV, the technology trend has a distinct and separate trajectory from NFV.
In the past couple of years, we have seen operators globally commercialize networks with SDN control of optical elements (Layers 0/1), Ethernet switches (Layer 2) and IP routers (Layer 3). For the most part in these carrier networks, SDN controls physical hardware elements, not virtualized software implementations running on COTS servers. Beyond CPE and access, performance requirements are typically too demanding for COTS, and, in the case of optics, there is no way to virtualize a photon. Successfully applying SDN to the physical world is critical for operators because physical elements make up the vast majority of their networks. Progress in the physical works is a good sign.
A promising -- though challenging -- next step is to begin applying SDN control across network layers, and particularly across the IP and optical layers. While operators are applying SDN to optical networks and IP networks today, they are applying it separately, such that the layers remain largely isolated islands of operation.
Such software-based integration is important to operators. In our survey, 68% of respondents reported that SDN-based IP and optical integration as at least "very important" for their next-generation networks, with 27% of the group identifying IP and optical integration as "critical." In fact, SDN-based IP and optical integration scored higher marks on the priority list than element disaggregation -- a more high-profile trend that was also a topic in our survey.
Not surprisingly, operators cited lack of standardization as a top barrier to commercial progress in integrating IP and optical control and management. Here, the physical elements story echoes the one playing out today in the virtualized world. Applications work well in tightly controlled environments -- within a data center, across a private data center network, or within a single vendor domain. But SDN's (and NFV's) greatest benefits will come when it can control complex network scenarios, across multiple domains, vendors and network layers. To master complexity, there is a lot work still to be done.
For more about our global carrier SDN survey, please see Charting the Path to Network Automation & Disaggregation: Carrier SDN Survey Analysis.
— Sterling Perrin, Principal Analyst, Heavy Reading